INDEX

Explanations

mentions of the term "SA" with high activation

oai_token-act-pair · gpt-3.5-turbo

references to the term "SA" which appears to indicate a specific organization or entity

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.1.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 MacArthur

-0.82

starter

-0.80

hower

-0.76

lace

-0.72

deck

-0.71

naire

-0.69

ships

-0.68

hai

-0.68

library

-0.65

mony

-0.64

POSITIVE LOGITS

SA

1.27

VE

1.16

ULT

0.99

FA

0.99

BIL

0.94

GE

0.92

VER

0.90

GA

0.90

KER

0.89

PA

0.89

Activations Density 0.004%

mentions of the term "SA" with high activation

references to the term "SA" which appears to indicate a specific organization or entity

No Comments

No Known Activations